Programming ML.NET

Chapter 8

Forecasting Tasks

“Complete chaos is impossible.”

—Theodore Motzkin, talking about Ramsey’s theory, 1951

In Chapter 7, where we discussed spotting anomalous values in data series, we introduced the concept of time series. As a reminder, a time series is a sequence of values captured at successive, ideally equidistant points in time. Therefore, a time series is a discrete (as opposed to continuous) collection of values. The wind speed values reported every 30 seconds by the anemometer installed on a specific wind turbine in a specific farm is a good example of a time series.

There are two types of information that can be extracted from a time series. One type of information is spikes and change points, namely anomalous values. A spike ...